And a floating point processing unit, while we are at it. PIC32s have them, and Intel has the IP to implement it. I really don't understand why they implemented the FPU on the CPUs, but "missed" to add that on the MPUs.
It's a specialty item found on upper end MCU offerings as most applications don't need it. For instance the lower end PIC32's and ARM's don't have it.
The D2000 is a very low end product. It won't be used in apps needing FP.
M4 and M7 ARMS which do have it are on the opposite end of the spectrum. Feature rich with gobs of memory and peripherals. They are also scary complex as well.
Look at the P2 - it's not designed for apps needing FP and DSP capabilities. It's quite capable, but it's targeted to more I/O oriented needs. It's more or less a front end chip that would be talking to a M7 or a Sitara on the back end.
And a floating point processing unit, while we are at it. PIC32s have them, and Intel has the IP to implement it. I really don't understand why they implemented the FPU on the CPUs, but "missed" to add that on the MPUs.
Only a small subset of Microcontrollers have FPU - yes, it is very easy to be in an armchair and say 'add a FPU, while we are at it', but there is a cost adder to that, and that is why sales of those MCUs without FPU, greatly outnumber those with.
If we look at more experienced players, like Freescale, their K10 series shows a FPU only on some models, and they choose to put this only on the very top end 512/1M Flash large parts, at prices 2x that of intel's.
I don't have armchairs at home, unfortunately. And yes, if Intel implements a true CPU core as they claim, I don't see the reason why not.
And a floating point processing unit, while we are at it. PIC32s have them, and Intel has the IP to implement it. I really don't understand why they implemented the FPU on the CPUs, but "missed" to add that on the MPUs.
It's a specialty item found on upper end MCU offerings as most applications don't need it. For instance the lower end PIC32's and ARM's don't have it.
The D2000 is a very low end product. It won't be used in apps needing FP.
M4 and M7 ARMS which do have it are on the opposite end of the spectrum. Feature rich with gobs of memory and peripherals. They are also scary complex as well.
Look at the P2 - it's not designed for apps needing FP and DSP capabilities. It's quite capable, but it's targeted to more I/O oriented needs. It's more or less a front end chip that would be talking to a M7 or a Sitara on the back end.
Hi Rod,
I thought that the D2000 as a higher-end product and a contestant to the ARM Cortex M4. These are quite cheap, and still have an FPU. Therefore, I don't see why the boast about. I mean, single core, 32MHz operation, it is not better than a 32-bit PIC. So, I think Intel doesn't have the egde there.
Well, as for the P2, it has CORDIC, and that suffices (very fast and even used in calculators). P1 doesn't support them in the hardware, and it is a bit of a downer for more specialized applications, but still does the job well in software, given the speed. In fact, I've tried to get away without C sqrt(), by using a fast integer sqrt version, and the new algorithm (from whatever source) was quite slower.
I thought that the D2000 as a higher-end product and a contestant to the ARM Cortex M4. These are quite cheap, and still have an FPU.
The detail that matters here, is not all M4 have the FPU, it is an optional add-on.
With 32KF and 8KR, and $2.53 and 40 pins, I would not really call that a 'higher-end' product either in 2016.
These days, 512~2M, is more higher end.
Perhaps a smarter comparison, is to find other MCUs at the 40 pin area, that can do ?
32*32 -> 64 and
64/32 -> 32q:32r
I thought that the D2000 as a higher-end product and a contestant to the ARM Cortex M4. These are quite cheap, and still have an FPU.
The detail that matters here, is not all M4 have the FPU, it is an optional add-on.
With 32KF and 8KR, and $2.53 and 40 pins, I would not really call that a 'higher-end' product either in 2016.
These days, 512~2M, is more higher end.
Perhaps a smarter comparison, is to find other MCUs at the 40 pin area, that can do ?
32*32 -> 64 and
64/32 -> 32q:32r
Hi jmg,
Your first point is reasonable. Your second point I don't understand, since it is cryptographic. Could you use other words?
Perhaps a smarter comparison, is to find other MCUs at the 40 pin area, that can do ?
32*32 -> 64 and
64/32 -> 32q:32r
Your first point is reasonable. Your second point I don't understand, since it is cryptographic. Could you use other words?
I was meaning the Maths operators, in D2000 I think they can do
32b * 32b to 64b result, and also
64b/32b to give 32b Quotient and 32b remainder.
That's quite good maths, for a 40p 32k device.
I think PIC32 can do the multiply, but the divide is vague, in some places mentions 64b, but never shows which pair of registers are needed for source.
Cortex I think has no Quotient and Remainder Div, but can do a extra multiply + subtract to get remainder.
It really should not be this hard to find decent opcode information on MicroControllers.
Comments
It's a specialty item found on upper end MCU offerings as most applications don't need it. For instance the lower end PIC32's and ARM's don't have it.
The D2000 is a very low end product. It won't be used in apps needing FP.
M4 and M7 ARMS which do have it are on the opposite end of the spectrum. Feature rich with gobs of memory and peripherals. They are also scary complex as well.
Look at the P2 - it's not designed for apps needing FP and DSP capabilities. It's quite capable, but it's targeted to more I/O oriented needs. It's more or less a front end chip that would be talking to a M7 or a Sitara on the back end.
I thought that the D2000 as a higher-end product and a contestant to the ARM Cortex M4. These are quite cheap, and still have an FPU. Therefore, I don't see why the boast about. I mean, single core, 32MHz operation, it is not better than a 32-bit PIC. So, I think Intel doesn't have the egde there.
Well, as for the P2, it has CORDIC, and that suffices (very fast and even used in calculators). P1 doesn't support them in the hardware, and it is a bit of a downer for more specialized applications, but still does the job well in software, given the speed. In fact, I've tried to get away without C sqrt(), by using a fast integer sqrt version, and the new algorithm (from whatever source) was quite slower.
Kind regards, Samuel Lourenço
With 32KF and 8KR, and $2.53 and 40 pins, I would not really call that a 'higher-end' product either in 2016.
These days, 512~2M, is more higher end.
Perhaps a smarter comparison, is to find other MCUs at the 40 pin area, that can do ?
32*32 -> 64 and
64/32 -> 32q:32r
Your first point is reasonable. Your second point I don't understand, since it is cryptographic. Could you use other words?
Kind regards, Samuel Lourenço
I was meaning the Maths operators, in D2000 I think they can do
32b * 32b to 64b result, and also
64b/32b to give 32b Quotient and 32b remainder.
That's quite good maths, for a 40p 32k device.
I think PIC32 can do the multiply, but the divide is vague, in some places mentions 64b, but never shows which pair of registers are needed for source.
Cortex I think has no Quotient and Remainder Div, but can do a extra multiply + subtract to get remainder.
It really should not be this hard to find decent opcode information on MicroControllers.